"Before the election, we also detected and took action on activity relating to hashtags that
have since been reported as manifestations of efforts to interfere with the 2016 election. For
example, our automated spam detection systems helped mitigate the impact of automated Tweets
promoting the #PodestaEmails hashtag, which originated with Wikileaks’ publication of
thousands of emails from the Clinton campaign chairman John Podesta’s Gmail account. The
core of the hashtag was propagated by Wikileaks, whose account sent out a series of 118 original
Tweets containing variants on the hashtag #PodestaEmails referencing the daily installments of
the emails released on the Wikileaks website. In the two months preceding the election, around
57,000 users posted approximately 426,000 unique Tweets containing variations of the #PodestaEmails hashtag. Approximately one quarter (25%) of those Tweets received internal
tags from our automation detection systems that hid them from searches. As described in greater
detail below, our systems detected and hid just under half (48%) of the Tweets relating to
variants of another notable hashtag, #DNCLeak, which concerned the disclosure of leaked emails
from the Democratic National Committee. These steps were part of our general efforts at the
time to fight automation and spam on our platform across all areas."
Relevant portion based on title. Part of an anti-spam effort.
I've never understood Twitter's "anti-spam" effort. The only "spam" I've ever seen on twitter is when someone I am following likes another tweet, or it tells me I should follow someone because someone I'm following follows them. I'm not interested in either of those suggestions. I'm only interested in what the exact account I'm following tweets or retweets, because I decided to follow it.
It seems to me that if the system worked as intended, I would see none of this mess because I'm not interested in following the bots. They can tweet away all they want, hundreds of thousand of them, to crickets.
I would venture to say that the "spam" is less of a problem for what users really want and more of a problem for their efforts that try to increase user retention/engagement and therefore ad rev.
Here's a tip: If you make a list of users and look at that, the list will contain only tweets and retweets of the members, in chronological order. (I.e., like how the main timeline used to work, before twitter messed it up.)
For a long time if I posted something that the general public would be interested in (say, containing "iPhone") I'd get @ mentioned by a spam account for a related scam ("free iPhone!") or an unrelated spam for a "dating" site by an account with an NSFW profile pic.
I just recently learned you can hide retweets from your timeline for an individual user. I don't think you can turn off the likes, but that would be a good addition.
Yep, me as well, I wish I could activate that globally for all people I follow at the same time. Retweets are the worse in terms of useless signaling. They also added a filter filtering tweets containing words. What I would like is the opposite, for user U, filter all tweets that do not contain word X,Y,Z so people stay on topic. I'd also like to see a record of the tweet that made me follow user U.
There is quite A LOT of that these platforms can do if really want to make the experience better and stop the flow of information that is completely irrelevant to the user.
I believe the "don't show retweets for x user" feature is limited to the web interface and is not available in the mobile/desktop apps, which is basically the only way I consume Twitter. Which is disappointing. I may be wrong here (I hope).
As a Twitter lover, it's been a real letdown to see them sacrifice objectivity and fairness at the altar of political correctness. The platform is so seriously broken by their attempts to censor things they deem unpleasant. I have tried my best to turn off all of their content filtering features, but they still sometimes censor replies to my own replies. You can sometimes see the actual number of replies to a tweet next to the reply button, but can not actually see them at all.
At the same time, they have not done anything to reduce certain types of spam, such as the repeated reply chains under the President's tweets, which can so easily be identified as spam by the fact that they are unrelated to the parent, and appear in the same order under a number of parent tweets.
It's a total mess.
Added: I went back to find an example of the spam chains to reply to Srslyjosh, but I'm pleasantly surprised that they seem to have finally eliminated them. I withdraw that specific criticism I guess.
This isn't about the right-wing dogwhistle of "political correctness" at all. It's about not letting networks of fake users (bots and/or individuals claiming to be someone they aren't) spread false information with no input from real people. There are plenty of real, unrestricted "politically incorrect" people on Twitter right now. The difference - as it pertains to this discussion - is that they are real people, not robot armies paid for by foreign adversaries.
Yes, some of them have been banned, and sometimes the reasons for those bans are dubious because Twitter does not consistently enforce their policies. But that is a separate issue from what's under discussion here.
Never have I heard the expression "dog whistle" more than in the last two years, always regarding politics, and always from liberals chastising conservatives. It's like their "in group" slang.
enforcement consistency is the point of the discussion, otherwise Twitter can proactively or retroactively label anything 'spam' without describing how they [or their algorithm] arrived at that conclusion.
They did not suppress those tags because of "political correctness" but rather used those as an example of tags associated with unusually high levels of spam.
It does raise an interesting idea... If you want to suppress and discredit something, just spin up a pile of bots to heavily promote that which you want to censor. Bonus points if you route it through VPNs in Russia or China or North Korea.
It was alleged that bot nets would try and dilute politically motivated trends by blasting out a hashtag that was one letter off- for example #podasta instead of #podesta.
Then 'real' users who started typing would autofill with the false one and the topic would lose steam quicker. Of course, without knowing Twitters proprietary algorithms, this is all just anecdotal.
I'm not sure that invalidates the reasoning behind nuking hashtags driven by bot farms, which is clearly a practice that negatively impacts the practice. What you're describing sounds like a case for better discerning between bots and real users.
Boy we got here quickly. Could you please take a look at the guidelines, which ask us not to post like this?
> Eschew flamebait. Don't introduce flamewar topics unless you have something genuinely new to say. Avoid unrelated controversies and generic tangents.
No one believes that banning Nazis from Twitter is going to magically make them go away, but it does remove a platform for them, or at least hinder their ability to use it.
I don't know, but I remember when hating Nazi's was as American as apple pie or baseball.
Call of Duty, Wolfenstein, Battlefield 1942 if I recall correctly; were all anti-Nazi and it was basically it's own genre of first person shooters.
Then everything moved towards terrorism related themes (Counterstrike starting that I'm guessing)... then political correctness just referred to it as "Opposing Force (OpFor)".
No, the pro-Trump ones are just as annoying. Regardless of the opinion the chains take, it is annoying that they have nothing to do with the specific tweet.
I noticed this during that time and reported on my observations [1]. I'm glad to see there is a followup inquiry, even if it's over a year later.
The core issue I had back then still seems unclear. When I noticed the oddity, the #DNCLeak hashtag was being hidden / censored (at 3 tweets / sec), and during the same time the pro-democrat hashtag #RNCinCLE was trending strongly (at 12 tweets / min).
In this document, is twitter saying it was simply their spam detection and prevention efforts that caused the #DNCLeak to be hidden (and not any ulterior or political motives)?
Note that, contrary to what the HN title implies, Twitter doesn't say it specifically targeted those hashtags or hid them entirely. The testimony mentions them as examples of tags that were associated with a significant number of bot/spam tweets.
Other than following the election very closely and observing, no I don't have evidence. Tags like #whichhillary were very popular at times yet not in the trending tags. If you caught a tag before the censors you could see it trending before they removed it. Again, just observation, sadly.
Are their trending hashtags included in the firehose data or otherwise recorded? Someone could go back and analyze it if they had the data.
edit: Here you go. Also google for it to find more links. Feb 2016, well before the primary was over.
Yeah, the current title seems pretty misleading / politically slanted. The actual content of the article is a lot more. "Twitter's Testimony before the Senate Judiciary, Subcommittee on Crime and Terrorism" or something similar might make more sense. It's a 15 page document of Twitter's testimony, not a news article about the contents of that testimony.
"In the two months preceding the election, around 57,000 users posted approximately 426,000 unique Tweets containing variations of the #PodestaEmails hashtag. Approximately one quarter (25%) of those Tweets received internal tags from our automation detection systems that hid them from searches. As described in greater detail below, our systems detected and hid just under half (48%) of the Tweets relating to variants of another notable hashtag, #DNCLeak, which concerned the disclosure of leaked emails from the Democratic National Committee. These steps were part of our general efforts at the time to fight automation and spam on our platform across all areas."
I'm interested to know how they determined half of all #DNCleaks tweets were spam and how that compares to other trending topics both political and apolitical.
Seems like they are hiding behind the anti-spam stance without detailing how their system determines that. Removing almost 50% of communication on a topic is a pretty dramatic step, especially during an election.
In my experience analyzing these sorts of things for my skunkworks project, the "source" (client or app) and too-quick-to-be-human retweet brigades are dead giveaways, and I'd imagine they're part of this algorithm. I also started tracking "visual hashes" of avatars (blurred and reduced to a 32x32 grid) in an effort to find re-used or stock profile pics, which has shown some promise as a spammer/bot/brigade signal. As well, existing known signals such as "user with default background color" or "no verified email address" could be part of it.
For example, IFTTT was recently shown to be used by a pro-Trump "retweet brigade" attempting to boost the "Uranium One" campaign with old "RT-style" retweets that look like new tweets but all have the same text. That group wasn't terribly smart, since they all looked the same, and were broadcast in relatively short order by hundreds of accounts.
I know these are guesses; Twitter likely won't ever share their secret sauce recipe.
Yes, but it would be an equally dramatic step to leave their normal anti-spam system in place, but then specifically exempt a particular hashtag from filtering.
But of course you're right; the core problem is that the filtering process itself is completely opaque and inscrutable. We just have to take Twitter's word that it doesn't introduce explicit or implicit biases.
So Twitters explanation seems to be "Posts related to #PodestaEmails and #DNCLeak appeared to be being generated by automated systems and were therefore handled as spam"
> Before the election, we also detected and took action on activity relating to hashtags that have since been reported as manifestations of efforts to interfere with the 2016 election. For example, our automated spam detection systems helped mitigate the impact of automated Tweets promoting the #PodestaEmails hashtag, which originated with Wikileaks’ publication of thousands of emails from the Clinton campaign chairman John Podesta’s Gmail account. The core of the hashtag was propagated by Wikileaks, whose account sent out a series of 118 original Tweets containing variants on the hashtag #PodestaEmails referencing the daily installments of the emails released on the Wikileaks website. In the two months preceding the election, around 57,000 users posted approximately 426,000 unique Tweets containing variations of the #PodestaEmails hashtag. Approximately one quarter (25%) of those Tweets received internal tags from our automation detection systems that hid them from searches. As described in greater detail below, our systems detected and hid just under half (48%) of the Tweets relating to variants of another notable hashtag, #DNCLeak, which concerned the disclosure of leaked emails from the Democratic National Committee. These steps were part of our general efforts at the time to fight automation and spam on our platform across all areas.
"The effect of social transmission on real-world voting was greater than the direct effect of the messages themselves, and nearly all the transmission occurred between ‘close friends’ who were more likely to have a face-to-face relationship. "
So people that already were most likely aligned ideologically.
'Sci-Hub is a website with over 62 million academic papers and articles available for direct download.[2] It bypasses publisher paywalls by allowing access through educational institution proxies.'
If they had taken similar action against the provably bot-ridden tags of some of the pro-Hillary hashtags... they would have mentioned it in an effort to appear fair and balanced.
It's a case of "the dog that didn't bark" showing up their actions: Twitter has shown they are biased.
More like "too simple to be believed"... and easily seen on Twitter, by examining tweeters of Hillary's tweets and noticing that for instance, 10 or more accounts that simultaneously retweeted all had precisely the same number of tweets, in the thousands, and less than 10 followers apiece.
More accessible, however, is simply to notice that the testimony was being given to a government where the House, Senate, and Presidency are all Republican. If Twitter could have shown their even headedness, they would have done so.
You can very easily find plenty of mainstream media news articles about Hillary's million plus fake followers, and you can look at election era tweets and retweets yourself.
Yeah, I'm sorry, you must have been confused when I said reports and stories are not the same as evidence. And then you linked to the daily mail of all things. And the most credible report you linked to was just completely unhelpful to the point you're trying to make.
Don't bother me with your unmitigated stupidity again.
Relevant portion based on title. Part of an anti-spam effort.